【Objective】 To investigate the feasibility of using surface electromyography （SEMG） for the detection of abnormal muscle response （AMR） in patients with hemifacial spasm （HFS）. 【Methods】 We retrospectively reviewed the clinical data of HFS patients who underwent microvascular decompression （MVD） in our hospital between June 2019 and December 2020. Patients who received both surface electrode （preoperative） and needle electrode （intraoperative） detection of AMR were included. SEMG recorded from two stimulation-recording sites， namely， zygomatic-mentalis and mandibular marginal-orbicularis oculi， was selected for analyzing the characteristics of AMR. The positive rates of AMR detected by these two kinds of electrodes were comprehensively compared. 【Results】 Totally 77 patients were included in this study. When detected with surface electrodes， the positive rate， latency and amplitude of AMR recorded at zygomatic-mentalis oculi were 90.9% （70/77）， （10.87±1.86） ms and （202.8±47.4） μV， and at mandibular marginal-orbicularis oculi were 92.2% （71/77）， （10.41±1.83） ms and （211.1±54.1） μV， respectively. AMR was detected in 74 patients （96.1%） with surface electrodes. There was no significant difference in positive rate， latency and amplitude of AMR between these two stimulation-recording methods. When detected with needle electrodes， the positive rate of AMR recorded at zygomatic-mentalis oculi was 98.7% （76/77）， which was significantly higher than the rate 89.6% （69/77） recorded at mandibular marginal-orbicularis oculi （P=0.016）. The latency and amplitude of AMR recorded at zygomatic-mentalis were （10.63±1.39） ms and （83.5±27.2） μV， and at mandibular marginal-orbicularis oculi were （10.31±1.18） ms and （58.6±21.4） μV. There was no significant difference in latency between the two stimulation-recording methods， but the amplitude recorded at mandibular marginal-orbicularis oculi was significantly lower （P=0.041）. AMR was detected in 76 patients （98.7%） with needle electrodes. There was no significant difference in the detection rate of AMR between surface electrodes and needle electrodes （P=0.500）， the results were moderately consistent （Kappa=0.490， P<0.001）. 【Conclusion】 The detection efficiency of surface electrodes for AMR is similar to that of needle electrode. With its non-invasive characteristic， the surface electrode can be routinely used for electrophysiological evaluation of HFS.